WW2 BAD-T III - BDAc AI Dogfight Tournament

[DoctorDavinci]

@Arvolder's  DRAWROF MkIII vs @MightyDarkStar's  Kestrel F. 1

 

[SuicidalInsanity]

@dundun92's P-57 vs @53miner53's WWII Fighter 1:

 

[Guest]

3 hours ago, SuicidalInsanity said:

@dundun92's P-57 vs @53miner53's WWII Fighter 1:

 

as expected.

[53miner53]

7 hours ago, dundun92 said:

as expected.

Yep

[SuicidalInsanity]

Now the proper video is uploaded, lets try this again.
@Eidahlil's Wasp Mk3.3 vs @aleksey444's AF 292D-Falconet:

 

Edited November 1, 2017 by SuicidalInsanity

[aleksey444]

Well, to quote @dundun92, that went "as expected".  I'm surprised my plane won even 1 out of 3.

Good job, @Eidahlil!

[Guest]

1 minute ago, aleksey444 said:

Well, to quote @dundun92, that went "as expected".  I'm surprised my plane won even 1 out of 3.

Good job, @Eidahlil!

Im surprised too... I AM DOOMED!

[aleksey444]

1 minute ago, dundun92 said:

Im surprised too... I AM DOOMED!

I think everybody is doomed.

Wasp can out-turn every other plane, and it's ridiculously accurate.

[Guest]

19 hours ago, aleksey444 said:

I think everybody is doomed.

Wasp can out-turn every other plane, and it's ridiculously accurate.

Remember the armor on @DoctorDavinci's Slap Chop... 

[aleksey444]

1 minute ago, dundun92 said:

Remember the armor on @DoctorDavinci's Slap Chop... 

It's not going to save him.

BDA bullets work by increasing a part's heat.  When heat exceeds the part's rating, the part explodes (or falls off...).  Most parts have 2000 heat max.  B9 wings are 2400, I believe.  The armor is 5000-5500 or so.  I'm working from memory, because I don't have KSP running at the moment.  Also, somehow, I believe it depends on the part's size (or mass) -- bigger can absorb more heat.  I don't have a very thorough understanding of it, unfortunately.  However, most armor parts are small and light.

You can press F3 in the middle of the battle to get a report of which part(s) got blown off.  I've seen temperatures of 12,000 or more, caused by bullets.

Essentially, armor means you can take 1-2 hits more than usual (and the usual is 1-2 hits itself).  It depends on the bullets, of course.  So, maybe it's not 1-2, but 3-4 or something.  But it's not a lot.

The Wasp is incredibly accurate.  Because of its large wings (probably -- I wish I could build planes like that), the Wasp doesn't shake -- it fires in a very straight line.  And it's firing time is set to several seconds, at least.

The Wasp is nimble and can turn on a dime.  Slap Chop is not as maneuverable.  From watching the video of it's first fight, it was comparable to Rolf.  And Rolf is slightly, slightly less maneuverable that my plane (it was my main sparring partner when I was building mine).  Wasp killed me easily.

So, my prediction is that the Wasp will tear Slap Chop to shreds .  It will just take 0.5 seconds longer than with other planes.

[Van Disaster]

Hispanos care not for armour - when I've run tests armoured craft have just lost like they didn't have any. The Wasp takes full advantage of "it's a game" - no pilot ever is going to hit other planes at over 1km with WW2 cannon & those big wings ain't happening IRL either at that weight. If you could get a plane to sit up high & drop down on it occasionally you'd make it stall out a lot & it'd be a sitting duck given how big it is, but I've yet to reliably manage that with the autopilot. It's roll rate isn't great either but with that turn rate it doesn't need to be.

I remember reading about the RAF trying to work out tactics in the 60s of how to fight WW2 prop fighters with Lightnings, because the Indonesians had some P-51s & the Borneo conflict was going on ( I think I would rather have been in a Hunter for that fight... ). They decided in the end that dropping down behind & underneath was the best way ( which is usually not a good idea ), which is just modified B&Z really - unfortunately we can't tell our pilots to do that when they're up against slow turnfighters

I suppose you could have just dropped on top of it & given it a cockpit full of twin afterburners

Edited November 2, 2017 by Van Disaster

[aleksey444]

53 minutes ago, Van Disaster said:

... those big wings ain't happening IRL either at that weight.

That's what I thought, too.

I've built fighters for other competitions before, although I only submitted one prior to this one.  And I've noticed that big (in fact, huge) wings help tremendously.  I don't know the right aerodynamic terms for it, but big wings "grip" or "bite into" the air better when turning, so the plane can turn faster and is more stable in a turn.  This might very well the the case in real life as well -- after all, FAR models real aerodynamics, but there's got to be a reason we don't build planes like that, probably because real-world materials are not strong enough.

Still, it's a game, and props to @Eidahlil for taking advantage of the game mechanics in a successful way.

[Eidahlil]

A few notes on the design of the Wasp, in case someone is interested

Turning: yes, the big wings help immensely with turning (wiki) - bigger wings allow either to turn at the same speed with a lower AoA, thus experiencing less drag and retaining more energy, or to turn faster at the same AoA - without losing more energy. The fast pitch turning is additionally increased by using a stabilator (all-moving tail), unlike, I think, all the other entries, and having a relatively narrow mass distribution along the length of the airplane. The AI pilot of the Wasp is tuned to turn very aggressively (steer factor = 24), however due to the large wings and low speeds the Wasp changes its' velocity vector quickly enough not to achieve a large AoA even with rapid pitch maneuvers .

The pilot does indeed stall the plane sometimes, as discussed previously somewhere in the thread, but the Wasp has a relatively high TWR so it recovers quickly. Regarding TWR - based on the introductory part of the first round of fights I believe the Wasp Mk3.3 is the lightest plane in the competition, while also being only one of two planes to use the most powerful Falcon engine (the other being the Deimos).

The large wings are not without drawbacks, however. They produce a lot of drag when not turning, so despite its' high TWR, the Wasp is usually moving at a lower velocity than the other planes.

Regarding realism, if I count right, the main wing area is about 45 m², so at a mass of 2.8 t, the wing loading is ~62 kg/m², which is fairly close to a light airplane (e.g. 1961 Piper Cherokee), but only half of, e.g., Supermarine Spitfire. Wiki of example wing loadings. So, uh, not sure how realistic that is?

Guns: The accuracy is achieved mainly through having stable flight. Stable flight is achieved through having large control surfaces - for a given control surface turning speed, a larger surface will achieve the desired correction faster, resulting in less fluctuation. The steer damping is also turned up all the way to 12.

There are a few other tricks, though: since the 20mm needs only a couple shots to do serious damage, the guns are spread out a bit in a kind of shotgun pattern, maximizing the probability of scoring a hit or two. The guns are also slightly tilted upwards to compensate for bullet drop and aerodynamic qualities of the airplane (i.e. if you see your plane shooting under the target all the time, tilt the guns up ).

Spoiler

Also, the guns are set to fire in very short bursts, the target scan interval is only half a second, which means the Wasp is likely to take advantage of all opportunities to fire. I am uncertain how much it helps, but the guns are also set to fire in salvo mode, instead of barrage, which might reduce shaking, not sure.

All that does not come without drawbacks, however - this firing pattern overheats the Hispanos quickly, often even in the initial exchange, leaving the Wasp vulnerable for a period after the initial joust. If they did not use the heavy 20mm bullets as ammo, I'd probably have gone with 4xMG151 instead of Hispanos, as they have superior ROF, muzzle velocity and do not overheat so quickly. But with more guns and higher ROF that would mean I'd have to take another box of the 20mm ammo, and that's 90kg which can be avoided.

Regarding realism, some random article I googled suggests that for burst of 10-12 bullets a WW2 .50 machine gun achieved a spread of about 1.5 m at a distance of 0.5 km or so, which would mean the spread is about 3 m at a distance of one kilometer. Sounds accurate enough for a bullet or two to hit an airplane in a few bursts. Could a pilot actually aim at that distance - I do not know.

[Van Disaster]

It's not just about wing loading, it's about structural integrity - at the G that craft pulls in turns you'd fold the wings into U shapes ( nearly half the craft mass is in the engine in the centre of the wings let's not forget ) if that was a real wing, and that's not bringing up stresses from speed, or flutter, or a myriad of stresses from rotations. The same reasons let us build crazy spaceplanes, so it's not a terrible thing. There was a fully aerobatic WW2 bomber called the Shorts Stirling ( so much so that it could out-turn fighters ) - just have a look at how monstrous it's wing was to deal with the stresses on that large wing & how terrible the rest of it's performance was thanks to that wing.

Guns-wise, cannon had rather lower muzzle velocity - but most of it would be the pilot trying to even see a plane over a km away through a basic gunsight, when they're trying to deal with combat maneuvers & air turbulence & looking round trying not to get bounced. Successful pilots got kills at 100m, not 1000m...

Edited November 2, 2017 by Van Disaster

[Guest]

12 hours ago, Eidahlil said:

A few notes on the design of the Wasp, in case someone is interested

Turning: yes, the big wings help immensely with turning (wiki) - bigger wings allow either to turn at the same speed with a lower AoA, thus experiencing less drag and retaining more energy, or to turn faster at the same AoA - without losing more energy. The fast pitch turning is additionally increased by using a stabilator (all-moving tail), unlike, I think, all the other entries, and having a relatively narrow mass distribution along the length of the airplane. The AI pilot of the Wasp is tuned to turn very aggressively (steer factor = 24), however due to the large wings and low speeds the Wasp changes its' velocity vector quickly enough not to achieve a large AoA even with rapid pitch maneuvers .

The pilot does indeed stall the plane sometimes, as discussed previously somewhere in the thread, but the Wasp has a relatively high TWR so it recovers quickly. Regarding TWR - based on the introductory part of the first round of fights I believe the Wasp Mk3.3 is the lightest plane in the competition, while also being only one of two planes to use the most powerful Falcon engine (the other being the Deimos).

The large wings are not without drawbacks, however. They produce a lot of drag when not turning, so despite its' high TWR, the Wasp is usually moving at a lower velocity than the other planes.

Regarding realism, if I count right, the main wing area is about 45 m², so at a mass of 2.8 t, the wing loading is ~62 kg/m², which is fairly close to a light airplane (e.g. 1961 Piper Cherokee), but only half of, e.g., Supermarine Spitfire. Wiki of example wing loadings. So, uh, not sure how realistic that is?

Guns: The accuracy is achieved mainly through having stable flight. Stable flight is achieved through having large control surfaces - for a given control surface turning speed, a larger surface will achieve the desired correction faster, resulting in less fluctuation. The steer damping is also turned up all the way to 12.

There are a few other tricks, though: since the 20mm needs only a couple shots to do serious damage, the guns are spread out a bit in a kind of shotgun pattern, maximizing the probability of scoring a hit or two. The guns are also slightly tilted upwards to compensate for bullet drop and aerodynamic qualities of the airplane (i.e. if you see your plane shooting under the target all the time, tilt the guns up ).

  Hide contents

Also, the guns are set to fire in very short bursts, the target scan interval is only half a second, which means the Wasp is likely to take advantage of all opportunities to fire. I am uncertain how much it helps, but the guns are also set to fire in salvo mode, instead of barrage, which might reduce shaking, not sure.

All that does not come without drawbacks, however - this firing pattern overheats the Hispanos quickly, often even in the initial exchange, leaving the Wasp vulnerable for a period after the initial joust. If they did not use the heavy 20mm bullets as ammo, I'd probably have gone with 4xMG151 instead of Hispanos, as they have superior ROF, muzzle velocity and do not overheat so quickly. But with more guns and higher ROF that would mean I'd have to take another box of the 20mm ammo, and that's 90kg which can be avoided.

Regarding realism, some random article I googled suggests that for burst of 10-12 bullets a WW2 .50 machine gun achieved a spread of about 1.5 m at a distance of 0.5 km or so, which would mean the spread is about 3 m at a distance of one kilometer. Sounds accurate enough for a bullet or two to hit an airplane in a few bursts. Could a pilot actually aim at that distance - I do not know.

How did you get the green lines in the SPH?

[Van Disaster]

1 hour ago, dundun92 said:

How did you get the green lines in the SPH?

Push F2.

@Eidahlilyou have a point about the guns - my plane was generally flying nose-down for a while as I tried out various amounts of incidence - I think I reset the guns after & forgot to redo them...

Edited November 2, 2017 by Van Disaster

[DoctorDavinci]

@tetryds's Elkow-6 vs @Van Disaster's D_A_C Deimos Mk X

 

[Van Disaster]

I do so love those flappy wings the Elkow looks like a 30s air-racer, great stuff. Where did you hire those suicidal pilots, Dr D??

The end of the first fight was a great piece of ( perhaps fortunate ) camera work.

Edited November 3, 2017 by Van Disaster

[tetryds]

56 minutes ago, Van Disaster said:

I do so love those flappy wings the Elkow looks like a 30s air-racer

Yeah, I knew it would underperform a lot, I lose about 40% of lift when turning just because they bend so much, but it didn't matter.

It also has a terrible time trying to aim, for some reason it's simply too bad at aiming.

 

Btw, why were people discussing realism on BAD-T?

It's about having an interesting framework for a battleground, which fortunately happens when you add pieces of realism into it, the original purpose of BAD-T was never to be "realistic" per se.

The wasp design is interesting, and greatly exceeded my first expectations, but it's also a big target.

Would probably get rekt very easily by a fast sniper like the Tytonid or a tighter turner like the SkyRebal.

 

Edit: if this was a replica tournament it would be boring as hell.

But anyone is free to make one and see how it turns out.

Edited November 3, 2017 by tetryds

[Van Disaster]

19 hours ago, tetryds said:

Yeah, I knew it would underperform a lot, I lose about 40% of lift when turning just because they bend so much, but it didn't matter.

It also has a terrible time trying to aim, for some reason it's simply too bad at aiming

Ouch, that's a bit of a hit. My plane can't aim either, and you can see how twitchy & quick to change nose angle it can be.

As for realism/replicas, while they're intersecting sets they're not the same - I see a discussion about two particular BDA & general structural game mechanics vs real life, which is the same for *any* BDA contest. Replicas don't have to be realistic, just look right ( or you'd never have half size replicas, or VW-powered kit cars, etc etc ). My plane isn't a replica, it's a teardrop fuselage with an elliptical wing and a deliberately small tail, shapes that were arrived at by thinking about function. That it very vaguely looks like a Spitfire if it was built by Hawkers is mostly because they were the only people of the era to really use elliptical wings & that's how that engine & cockpit combo work together in that paint. The only attempts at replicas I've built are the jets, and that was just to try and limit myself to shapes that didn't have any idea about everything we learned since then or they'd look like something out of the 60s. I probably wouldn't enter a replica contest because I'm terrible at replica props. I *did* pay attention to the way the plane looks so it did look like something which might have flown IRL, so if you want to call that a replica then fine - but it really it isn't, it's just a little aesthetic consideration.

On the other hand having to really think hard about how much wing to put on sounds an interesting engineering challenge. Having to get in much closer for gun kills would be nice for running your own fights, but the poor people videoing would have a hell of a workload increase...

I have doubts the SkyRebal would win that one - Tytonid, I don't know.

Edited November 3, 2017 by Van Disaster

[aleksey444]

3 hours ago, tetryds said:

The wasp design is interesting, and greatly exceeded my first expectations, but it's also a big target.

Would probably get rekt very easily by a fast sniper like the Tytonid or a tighter turner like the SkyRebal.

 

1 hour ago, Van Disaster said:

I have doubts the SkyRebal would win that one - Tytonid, I don't know.

 

Can we not find out?

[SuicidalInsanity]

Wasn't difficult to port the SkyRebal and Tytonid to 1.3, so I do know how they perform against the Wasp; maybe I'll throw a bonus video together after the tournament is done.

 

[SuicidalInsanity]

Some more AI weirdness with this fight.

@Bob_Saget54's Falcon Mk XII vs my IA-22 Stribog VII:

 

Edited November 4, 2017 by SuicidalInsanity

[Van Disaster]

There's a plane that can shoot straight!

As I thought, I reset the guns on my craft & forgot to retune them...

Spoiler

And, wierd how it goes - in testing my craft beats the Falcon a great deal of the time, but it lost to the Stribog & that last fight wasn't a particularily close. I wonder if there's some environmental quirkiness.

 

Edited November 4, 2017 by Van Disaster

[tetryds]

Nice battle, lots of turnfighting.

Spoiler

Yeah, the AI was very weird, the Stribog seemed to be derping hard.

The minimum altitude was probably too high, most of the time it just kept going below it, then climbing over the target altitude (it seems), kept doing that and forgot to fight.

I would personally rather if the battle was the top priority, instead of obeying desired altitude parameters.